Multi-satellite on-orbit service dynamic increment greedy assignment method based on resource balancing strategy

Abstract

The invention relates to a multi-satellite on-orbit service dynamic increment greedy assignment method based on a resource balancing strategy. The method comprises the following steps: recursively pushing a service satellite and a target satellite to an epoch moment when a task starts; constructing a circular orbit rendezvous cost model to estimate a rendezvous cost matrix between the service satellite and the target satellite; constructing a target function of a task planning problem model based on an overall equilibrium distribution strategy and the rendezvous cost matrix, and designing constraint conditions of the task planning problem model according to task starting and ending time and information of each service satellite; constructing a task planning problem model by using the target function and the constraint condition; designing a dynamic increment greedy algorithm to solve the task planning problem model to obtain an assignment result of each service satellite; and designing an access sequence and rendezvous time for an assignment result of each service star according to a dynamic planning algorithm to obtain a final task planning scheme of each service star. By adopting the method, the task planning efficiency can be improved while the solving precision is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of spacecraft mission planning, and particularly to a dynamic incremental greedy assignment method for multi-satellite on-orbit service based on a resource balancing strategy. Background Technique

[0002] With the gradual increase in the number of space-based control platforms deployed in orbit, the on-orbit control task mode of spacecraft will develop from the traditional "one-to-one" mode to "one-to-many" and "many-to-many" modes, making the difficulty of on-orbit control task planning increase significantly. It is necessary to establish an efficient task planning method for "many-to-many" space-based platforms. The space rendezvous mission planning problem mainly designs the rendezvous sequence and orbital maneuver plan, and allocates time for each sub-task, etc. For the space orbit maneuver problem, some people have established a Hohmann-Lambert four-pulse rendezvous model for the GEO on-orbit service problem, and used Q-learning to solve the satellite service sequence planning. Some people have compared and analyzed four orbital rendezvous strategies, namely Hohmann transfer, maneuver drift, radial maneuver, and Lambert transfer, for the GEO satellite on-orbit service problem, and then used a genetic algorithm to optimize the many-to-many on-orbit service task planning problem. At the same time, the problem of designing the rendezvous sequence of multiple spacecraft is a problem with multiple variables, multiple constraints, and multiple objective functions. The design variables are both discrete and continuous, and this kind of problem can be regarded as a mixed integer nonlinear programming (MILP) problem. For this kind of problem, some people have obtained the space debris cleaning sequence by grouping the target points and combining tree search and ant colony algorithm. Some people have proposed a greedy search and multi-round planning method for multi-objective inspection task planning under constraints such as illumination and relative position, and also achieved rendezvous with the target point by applying in-plane pulse maneuvers. Some people have used deep Q-learning to solve the on-orbit service resource allocation problem, considering the composite service mode of "one-to-many" and "many-to-one" and taking into account the comprehensive benefits of resource allocation and the overall energy consumption efficiency.

[0003] However, the disadvantages of the current "many-to-many" task planning method are as follows: (1) The solution of the rendezvous sequence and the optimization of the maneuver trajectory in the "many-to-many" task planning are nested problems. The double-layer nested optimization structure is a very time-consuming and complex problem, and as the task scale increases, the solution time will show an exponential growth trend, making it still a certain challenge to quickly plan the "many-to-many" on-orbit control task, and it is difficult to meet the real-time requirement of on-orbit mission planning and the constraint of limited inter-satellite computing resources. (2) The current "many-to-many" task planning mostly uses linear integer programming methods to solve. This method can ensure finding the global optimal solution, but the computational complexity is exponential, and it is usually only applicable to small and medium-scale problems ( ). If the problem scale is large, more efficient solvers such as Gurobi or CPLEX may be required for support, but the solution efficiency is relatively low. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a multi-satellite on-orbit service dynamic incremental greedy assignment method based on a resource balance strategy that can improve the efficiency of task planning while ensuring the solution accuracy.

[0005] A multi-satellite on-orbit service dynamic incremental greedy assignment method based on a resource balance strategy, the method comprising: Obtain the start and end times of the task, the service satellite and the target satellite; recursively push the service satellite and the target satellite to the epoch moment when the task starts; construct a circular orbit rendezvous cost model to estimate the rendezvous cost matrix between the service satellite and the target satellite; Construct the objective function of the task planning problem model based on the overall balance allocation strategy and the rendezvous cost matrix, and design the constraint conditions of the task planning problem model according to the start and end times of the task and the information of each service satellite; use the objective function and the constraint conditions to construct the task planning problem model; Design a dynamic incremental greedy algorithm to solve the task planning problem model to obtain the assignment results of each service satellite; design the access sequence and rendezvous time according to the assignment results of each service satellite by the dynamic programming algorithm to obtain the final task planning scheme of each service satellite.

[0006] The above multi-satellite on-orbit service dynamic incremental greedy assignment method based on the resource balancing strategy first constructs a circular orbit rendezvous cost model to estimate the rendezvous cost. The circular orbit rendezvous cost model provides an accurate cost basis for task allocation, ensures the solution accuracy, and simplifies the calculation process. Based on the overall balanced allocation strategy and the rendezvous cost matrix, an objective function is constructed, and the constraint conditions are designed according to the task start and end times and the service satellite information, making the task planning problem model conform to the actual situation and ensuring the solution accuracy. Reasonable constraint conditions narrow the search space, reduce the calculation amount, and improve the task planning efficiency. Then, based on the overall balance, the task allocation is carried out. Finally, the access sequence and rendezvous time planning are carried out for each service satellite, avoiding the repeated invocation of the inner optimization module, significantly improving the planning efficiency, and not affecting the solution accuracy at the same time. The dynamic incremental greedy algorithm is designed to formulate the allocation rules and dynamically update the results according to the overall balance strategy, quickly complete the task allocation while ensuring a certain accuracy, and greatly improve the efficiency compared with the traditional linear integer programming method. Finally, the dynamic programming algorithm is used to solve the access sequence and rendezvous time of the service satellite, decompose the complex problem into sub-problems, and avoid repeated calculations by storing the solutions of the sub-problems, which not only ensures the calculation of accurate action plans but also improves the algorithm efficiency. This application adopts a framework that decouples task allocation and sequence planning, avoiding the repeated invocation of the inner optimization module. Based on the characteristics of task planning, the maneuvering scheme between adjacent target satellites is calculated according to the specified rendezvous strategy at the task level, avoiding the iterative solution of the innermost "one-to-one" maneuvering scheme, which is beneficial to further improve the generation efficiency of the action plan. Description of the Drawings

[0007] Figure 1 It is a schematic flowchart of a multi-satellite on-orbit service dynamic incremental greedy assignment method based on the resource balancing strategy in an embodiment; Figure 2 It is a schematic diagram of coplanar resection in an embodiment; Figure 3 It is a flowchart of the overall balanced allocation algorithm in an embodiment. Detailed Embodiments

[0008] In order to make the objectives, technical solutions, and advantages of this application clearer, the following further elaborates on this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0009] In one embodiment, as Figure 1 shown, a multi-satellite on-orbit service dynamic incremental greedy assignment method based on the resource balancing strategy is provided, including the following steps: Step 102: Obtain the start and end times of the task, the service star, and the target star; extrapolate the service star and the target star to the epoch moment when the task starts; construct a circular orbit rendezvous cost model to estimate the rendezvous cost matrix between the service star and the target star.

[0010] This application combines the characteristics of the "many-to-many" planning problem and adopts a task planning framework that decouples task allocation and sequence planning. In the whole process, first construct a circular orbit rendezvous cost model to estimate the rendezvous cost, then perform task allocation based on overall equilibrium (solved by the dynamic incremental greedy algorithm), and then plan the access sequence and rendezvous time for each service star (using the dynamic programming algorithm). This decoupling avoids the double-layer nested structure of rendezvous sequence solving and maneuver trajectory optimization in the traditional nested "many-to-many" task planning framework, prevents repeated calls to the inner optimization module, reduces the computational complexity and time consumption, and improves the task planning efficiency without affecting the solution accuracy. Because in the traditional nested structure, as the task scale increases, the solution time increases exponentially, while after decoupling, this complex nested relationship is broken, making the calculation process more concise and efficient.

[0011] Estimate the transfer cost between each pair of service stars and target stars according to the circular orbit rendezvous cost, which serves as the basis for fuel estimation in the outer task allocation. According to orbital dynamics, the circular orbit rendezvous cost model is divided into two models: coplanar phasing rendezvous and non-coplanar phasing rendezvous. Take the coplanar phasing rendezvous model as an example to illustrate its detailed scheme, as Figure 2 shown, when the target and the service star are on the same circular orbit, and the service star trails the target at a certain angle , in order to perform rendezvous, the service star must enter a lower orbit with a shorter period to eliminate the initial phase difference.

[0012] When the service star rotates one circle on the phasing orbit, the phase of the target star's motion is an integer multiple of the number of circles minus the initial phase angle , then the phasing time is: (1) In the formula, is the number of motion circles of the target star, and the initial phase angle carries a positive or negative sign, is the orbital angular velocity of the target star.

[0013] Since the service star needs to run an integer multiple of circles along the phasing orbit when completing the rendezvous, its semi-major axis is: (2) Since the research object is a circular orbit with tangential pulses applied, the velocity increments applied at different orbital radii are different. During the first maneuver, the service star must decelerate to enter the phasing orbit, and its velocity increment is (3) Total velocity increment is twice the initial velocity pulse: (4) The rendezvous cost matrix between the service satellite and the target satellite is estimated by constructing a circular orbit rendezvous cost model. The use of this model provides a relatively accurate cost basis for subsequent task allocation, enabling more reasonable task allocation and ensuring the solution accuracy. At the same time, subsequent calculations and planning are carried out based on this model, avoiding increased computational complexity and decreased accuracy caused by using overly complex or inaccurate models.

[0014] Step 104: Based on the overall balanced allocation strategy and the rendezvous cost matrix, construct the objective function of the task planning problem model, and design the constraint conditions of the task planning problem model according to the task start and end times and the information of each service satellite; use the objective function and constraint conditions to construct the task planning problem model.

[0015] Based on the overall balanced allocation strategy and the rendezvous cost matrix, construct the objective function of the task planning problem model, and design the constraint conditions of the task planning problem model according to the task start and end times and the information of each service satellite, making the task planning problem model more in line with the actual situation, so that more accurate results can be obtained during the solution process, ensuring the solution accuracy. At the same time, reasonable constraint conditions avoid unnecessary search spaces, reduce the computational complexity, and improve the efficiency of task planning. Among them, the overall balanced allocation strategy is as Figure 3 shown.

[0016] Step 106: Design a dynamic incremental greedy algorithm to solve the task planning problem model to obtain the assignment results of each service satellite; according to the assignment results of each service satellite by the dynamic programming algorithm, design the access sequence and rendezvous time to obtain the final task planning scheme of each service satellite.

[0017] Design the rendezvous cost matrix between the service satellite and the target satellite, design a dynamic incremental greedy algorithm for the mixed integer programming problem of task allocation, design the corresponding "service satellite - target satellite" allocation rules for the overall balanced allocation strategy, generate the corresponding task allocation scheme, and use the dynamic update mechanism to update the current allocation results for the next round of allocation, finally clarifying which target satellites each service satellite needs to control. While ensuring a certain solution accuracy, it can quickly perform task allocation. Compared with the traditional linear integer programming method, the computational complexity is exponential and only applicable to small and medium-sized problems, greatly improving the efficiency of task allocation and thus the efficiency of the entire task planning.

[0018] The dynamic programming algorithm is used to solve the access sequence and rendezvous time between different tasks of each service satellite, decomposing complex problems into simpler sub-problems and avoiding repeated calculations by storing the solutions of sub-problems. The order of the service satellite accessing the target satellite is modeled as a discrete non-linear system varying with time, and the velocity increment and transfer time of this maneuver are obtained through the above phasing maneuver calculation method. Combining the weighted coefficient input by the program, the cost of this level is characterized as a dynamic function, so as to solve the access order and maneuver time according to the above dynamic programming algorithm, and finally form the action plan of each service satellite. In this way, the action plan of each service satellite can be accurately calculated, ensuring the solution accuracy. At the same time, the feature of avoiding repeated calculations improves the efficiency of the algorithm, enabling quick results to be obtained when designing the access sequence and rendezvous time, and improving the efficiency of mission planning.

[0019] The above multi-satellite on-orbit service dynamic incremental greedy assignment method based on the resource balance strategy first constructs a circular orbit rendezvous cost model to estimate the rendezvous cost. The circular orbit rendezvous cost model provides an accurate cost basis for task allocation, ensuring the solution accuracy and simplifying the calculation process. Based on the overall balance allocation strategy and the rendezvous cost matrix, the objective function is constructed, and the constraint conditions are designed according to the start and end times of the tasks and the service satellite information, making the task planning problem model conform to the actual situation and ensuring the solution accuracy. Reasonable constraint conditions narrow the search space, reduce the calculation amount, and improve the efficiency of task planning. Then, based on the overall balance, task allocation is carried out. Finally, the access sequence and rendezvous time planning are carried out for each service satellite, avoiding repeated calls to the inner optimization module, significantly improving the planning efficiency, and at the same time not affecting the solution accuracy. The dynamic incremental greedy algorithm is designed to formulate the allocation rules according to the overall balance strategy and dynamically update the results, quickly completing the task allocation while ensuring a certain accuracy, and greatly improving the efficiency compared with the traditional linear integer programming method. Finally, the dynamic programming algorithm is used to solve the access sequence and rendezvous time of the service satellite, decomposing complex problems into sub-problems and avoiding repeated calculations by storing the solutions of sub-problems, ensuring both the calculation of accurate action plans and the improvement of algorithm efficiency. This application adopts a framework of decoupling task allocation and sequence planning, avoiding repeated calls to the inner optimization module, based on the characteristics of mission planning, calculating the maneuver plan between adjacent target satellites according to the specified rendezvous strategy at the task level, and avoiding iterative solution of the innermost "one-to-one" maneuver plan, which is beneficial to further improving the generation efficiency of action plans.

[0020] In one embodiment, the task planning problem model is constructed using the objective function and constraint conditions as: ; ; ; ; Among them, represents a decision variable, represents the rendezvous cost matrix, represents the serial number of the servicing satellite, represents the serial number of the target satellite, m represents the total number of servicing satellites, n represents the total number of target satellites, is the weight coefficient, represents the matrix element of the rendezvous cost matrix, represents the total velocity increment, represents the servicing satellite The time of the last rendezvous, represents the start and end time of the mission.

[0021] In a specific embodiment, based on the rendezvous cost matrix between the servicing satellite and the target satellite, a greedy algorithm based on dynamic increment is designed, and a corresponding "servicing satellite - target satellite" allocation rule is designed for the overall balanced allocation strategy to generate a corresponding task allocation plan. The overall balanced allocation strategy means that within the given start and end time, the tasks of the servicing satellites are assigned so that the servicing satellites can complete as many rendezvous tasks as possible, and the load of each servicing satellite is balanced, that is, the mean square deviation of the remaining fuel of each servicing satellite is the smallest.

[0022] Existing servicing satellites and target satellites, whose orbital elements and epoch times are all known, and the th servicing satellite can provide a total velocity increment of . Each target satellite can be assigned to at most one servicing satellite (not being assigned is allowed), and each servicing satellite can provide services for multiple target satellites in sequence. It is now required to assign the tasks of the servicing satellites within the given start and end time so that as many rendezvous tasks as possible are completed and the total velocity increment is the least.

[0023] First, construct the rendezvous cost matrix between the servicing satellite and the target satellite. Assume that both the servicing satellite and the target satellite are operating near the geostationary orbit. By using the circular orbit rendezvous cost estimation method introduced above, the velocity increment cost matrix and the maneuver time cost matrix of the servicing satellite rendezvousing with the target satellite can be obtained in the current state. Both of them are dimensional matrices. Taking the velocity increment cost matrix as an example, its matrix elements are as follows: (5) Among them, the matrix element represents the velocity increment of the servicing satellite rendezvousing with the target satellite .

[0024] Secondly, the 0-1 integer programming method is used to model the mission planning problem, and its decision variable is a Boolean variable , where: represents the target satellite assigned to the service satellite ; represents non-assignment.

[0025] The objective function of the integer programming model includes two: one is to complete as many rendezvous missions as possible, that is (6) The other is to minimize the velocity increment consumed to complete the mission, that is (7) It should be noted that the above rendezvous cost may be dynamically adjusted during the mission assignment process. Usually, the above two objective functions can also be integrated into one objective function in the form of a weighted sum, that is: (8) In the formula is the weight coefficient.

[0026] The constraint conditions of this integer programming model include three: one is that each target satellite can be assigned to at most one service satellite, that is: (9) The second is that the total velocity increment of each service satellite is limited, that is: (10) The third is that the final rendezvous time of each service satellite cannot exceed the start and end time of the mission, that is: .

[0027] In one of the embodiments, a dynamic increment greedy algorithm is designed to solve the mission planning problem model, and the assignment results of each service satellite are obtained, including: Sort the target satellites, calculate the set of assignable service satellites of the target satellites, define the priority of the target satellites, and sort the target satellites according to the priority; design an equilibrium index to traverse the sorted service satellites, select a suitable service satellite to be assigned to the current round of target satellites and update the remaining velocity increment of the service satellite and the assignment matrix between the service satellite and the target satellite , make the corresponding element , which is convenient for task assignment to the unassigned target satellites in the next round; At the beginning of the next round of allocation, it is necessary to update the orbits of the available service stars and the target stars to be allocated according to the new starting time, combined with the results of the previous round of allocation, and then calculate a new rendezvous cost matrix for this round of allocation. The start time of this round of allocation is set to the longest rendezvous time in the "service star - target star" combinations paired in the previous round. After completing each round of iteration in turn, the assignment results of each service star are obtained.

[0028] In a specific embodiment, a heuristic method based on a greedy strategy is designed to allocate tasks while satisfying resource constraints, including several stages as follows: i. Stage 1: Target star sorting: The main task of this stage is to preferentially allocate those target stars with low rendezvous costs and few available service stars to avoid being unable to complete due to resource competition. Its main steps include: 1) Calculate the set of available service stars for the target stars: For each target star , according to the velocity increment cost matrix filter out the service star set that satisfies , and calculate its quantity .

[0029] 2) Define the target star priority: Mainly consider the number of available service stars : The fewer the number of available service stars, the higher the priority (to avoid having no service stars available for arrangement later). Define the priority score of the th target star as: (11) In the formula to prevent the denominator from being 0.

[0030] 3) Sort the target stars by priority: Sort the target star list in ascending order according to , and preferentially process tasks with low scores.

[0031] ii. Stage 2: Service star allocation: The main task of this stage is to select appropriate service stars to be allocated to the current target star according to the balance index. Its main steps include: 1) Traverse the sorted target stars: In this section, the balance index is designed to minimize the difference in the remaining resources of all service stars after assignment: (12) In the formula, is the variance of the remaining resources of all service stars.

[0032] 2) Execute the allocation and update the available resources of the service star: Allocate the target star to the service star , and update the remaining velocity increment of the service star: (13) 3) Mark the allocated target star: Update the allocation matrix between the service star and the target star , making the corresponding element , which is convenient for task allocation of unallocated target stars in the next round.

[0033] iii. Dynamic update mechanism: Different from the static task allocation problem, in the task allocation problem between the service star and the target star, the rendezvous cost matrix between the available service stars and the unallocated target stars will change with the implementation of the allocated tasks. Assume that in the first-step decision, target stars have been allocated. Due to the greedy strategy set in the above-mentioned second stage being "select the service star with sufficient remaining resources, the lowest cost, and the target star that has not been allocated yet ", there must be service stars that have been assigned tasks, and the remaining target stars need to be allocated in the second-step decision.

[0034] At the beginning of the second-round allocation, it is necessary to update the orbits of the available service stars (with remaining resources) and the target stars to be allocated according to the new starting time, combined with the results of the previous-round allocation, and then calculate a new rendezvous cost matrix for the second-step allocation. For simplicity, the start time of the second-round allocation is set to the longest rendezvous time in the "service star - target star" combinations paired in the first round. After completing each round of iteration in turn, the assignment results of each service star are obtained.

[0035] In one embodiment, calculate the set of assignable service stars for the target star, including: For each target star , filter out the set of assignable service stars that satisfy according to the velocity increment cost matrix , and calculate its quantity ; where represents the total velocity increment.

[0036] In one embodiment, define the target star priority as ; where , represents the serial number of the target star, Represents the rendezvous cost matrix, i Represents the serial number of the service star.

[0037] In one embodiment, the designed balance index is

[0038] Wherein, Is the variance of the remaining resources of all service stars, i Represents the serial number of the service star, m Represents the total number of service stars, Represents the set of assignable service stars, Represents the average value of the remaining resource amounts of all service stars.

[0039] In one embodiment, according to the dynamic programming algorithm, the access sequence and rendezvous time are designed for the assignment result of each service star, and the final mission planning scheme for each service star is obtained, including: Model the order of service stars visiting the target star as a discrete nonlinear system that changes over time. Obtain the velocity increment and transfer time of the current round of maneuver according to the phasing maneuver calculation method. Combine the weighting coefficient input by the program to characterize the cost of this level as a dynamic function, and then solve the access order and maneuver time according to the dynamic programming algorithm to obtain the final mission planning scheme for each service star.

[0040] In one embodiment, model the order of service stars visiting the target star as a discrete nonlinear system that changes over time, including: The state equation for modeling the order of service stars visiting the target star as a discrete nonlinear system that changes over time is ; Wherein, define the system time As the th target star visited by the service star, and the state of the system time is the number of the th target star visited. The control quantity at time includes the number of the next target star to be visited and the number of turns of the phasing maneuver, expressed as .

[0041] In one embodiment, obtain the velocity increment and transfer time of the current round of maneuver according to the phasing maneuver calculation method, and combine the weighting coefficient input by the program to characterize the cost of this level as a dynamic function, including: Obtain the velocity increment and transfer time of the current round of maneuver according to the phasing maneuver calculation method, and combine the weighting coefficient input by the program to characterize the cost of this level as a dynamic function as ; wherein, represents the target star number where the current service star is located determined according to the state quantity, represents the number of the next target star accessed by the service star determined according to the control quantity, represents the number of turns of the phasing maneuver.

[0042] In a specific embodiment, the access sequence and the maneuver time planning problem can be regarded as a multi-level decision-making process. In the one-to-many access sequence and maneuver time planning problem of the present application, the order of the target stars accessed by the service star can be regarded as the change of the state of a discrete non-linear system over time. Define the system time as the th target star accessed by the service star, and the state of the system time is the number of the th target star accessed. The control quantity at time includes the number of the next target star to be accessed and the number of turns of the phasing maneuver, expressed as . Therefore, the system state equation can be expressed as .

[0043] Since there is no regulation on accessing a specific target star at the last time, the terminal value term in the cost function can take 0. The cost of each level can be determined by the state quantity and the control quantity at the current time. Specifically: determine the target star number where the current service star is located according to the state quantity, determine the number of the next target star accessed by the service star and the number of turns of the phasing maneuver according to the control quantity. The state of the target star corresponding to the number can be determined according to the program input, and the speed increment and transfer time of this maneuver can be obtained by the phasing maneuver calculation method. Combining with the weighting coefficient input by the program, it is converted into the cost of this level. Represent the above process as .

[0044] In one of the embodiments, the performance of the present application is verified by means of numerical simulation verification. Taking the many-to-many task allocation of 6 service stars to 12 target stars as an example, the access allocation results and access times are planned. It is known that the task start time is 0:10:04 on March 28, 2025, and the task end time is 0:10:04 on November 28, 2025. Other specific parameters such as the six orbital elements of the service stars and target stars at the initial moment, the propellant mass, and the platform priority are shown in Tables 1 and 2 below: Table 1

[0045] Table 2

[0046] Assume that the weight coefficients of the two indicators of the most fuel-efficient and the most time-efficient are 0.8 and 0.2 respectively. Then, when running the developed many-to-many task allocation program and adopting the balanced allocation strategy based on the two-stage greedy algorithm, the obtained many-to-many task allocation results are as follows: Table 3

[0047] As shown in Table 3, the task allocation results of each service star obtained by using the dynamic incremental greedy algorithm based on the overall balance strategy are shown. For example, the access order of service star 1 is target star 9 → target star 8, and the corresponding access times are 7.1342e+04 s and 1.6818e+05 s after the start time.

[0048] It should be understood that although Figure 1 the steps in the flowchart of Figure 1 are shown in sequence according to the indication of the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover,

[0049] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0050] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for dynamic incremental greedy assignment of multi-satellite on-orbit services based on resource balancing strategy, characterized in that: The method comprises: Obtain the mission start and end time, service satellite and target satellite; recursively calculate the service satellite and target satellite to the epoch time when the mission starts; construct a circular orbit rendezvous cost model to estimate the rendezvous cost matrix between the service satellite and the target satellite; The objective function of the mission planning problem model is constructed based on the overall equilibrium allocation strategy and the rendezvous cost matrix, and the constraint conditions of the mission planning problem model are designed according to the mission start and end time and the information of each service satellite; the mission planning problem model is constructed using the objective function and the constraint conditions; A dynamic incremental greedy algorithm is designed to solve the task planning problem model to obtain the assignment result of each service star; according to the dynamic programming algorithm, the access sequence and rendezvous time are designed for the assignment result of each service star to obtain the final task planning plan for each service star.

2. The method according to claim 1, characterized in that The task planning problem model is constructed using the objective function and constraints: in, represents the decision variable, represents the intersection cost matrix, i Indicates the serial number of the service star. Indicates the target star number, m Indicates the total number of service stars, n represents the total number of target stars, is the weight coefficient, represents the matrix elements of the intersection cost matrix, represents the total speed increment, Service Star i The last meeting time, Indicates the start and end time of the task.

3. The method according to claim 1, characterized in that A dynamic incremental greedy algorithm is designed to solve the task planning problem model and obtain the assignment results of each service star, including: Sort the target stars, calculate the set of assignable service stars for the target stars, define the priority of the target stars, and sort the target stars by priority; design a balanced index to traverse the sorted service stars, select appropriate service stars to allocate to the target stars of the current round, and update the remaining speed increment of the service stars and the allocation matrix between the service stars and the target stars , let the corresponding elements , which is convenient for the next round of task assignment to the unassigned target stars; At the beginning of the next round of allocation, it is necessary to update the orbits of the available service stars and the target stars to be allocated according to the new starting time and the results of the previous round of allocation, and then calculate the new rendezvous cost matrix for this round of allocation. The start time of this round of allocation is set to the longest rendezvous time in the "service star-target star" combination paired in the previous round. After completing each round of iteration, the assignment result of each service star is obtained.

4. The method according to claim 3, characterized in that Calculate the set of service stars that can be allocated to the target star, including: For each target star , according to the speed increment cost matrix Filter out the satisfied The set of allocable service stars , and calculate their number ;in, Indicates the total velocity increment.

5. The method according to claim 4, characterized in that Define the target star priority as in, , Indicates the target star number, represents the intersection cost matrix, i Indicates the serial number of the service star.

6. The method according to claim 3, characterized in that The design equilibrium index is in, is the variance of the remaining resources of all service stars, i Indicates the serial number of the service star. m Indicates the total number of service stars, represents the set of assignable service stars, Represents the average value of remaining resources of all service stars.

7. The method according to claim 1, characterized in that According to the dynamic programming algorithm, the visit sequence and rendezvous time are designed for the assignment results of each service star, and the final mission planning plan for each service star is obtained, including: The order in which the service satellite visits the target satellite is modeled as a discrete nonlinear system that changes with time. The speed increment and transfer time of the current round of maneuvers are obtained according to the phase adjustment motion calculation method. The cost of this level is represented as a dynamic function combined with the weighted coefficient input by the program. The visit order and maneuvering time are solved according to the dynamic programming algorithm to obtain the final mission planning plan for each service satellite.

8. The method according to claim 7, characterized in that The order of the service satellite visiting the target satellite is modeled as a discrete nonlinear system that varies with time, including: The order of service satellite visiting target satellite is modeled as a discrete nonlinear system that varies with time: Among them, define the system time Visited by the service star target star, system time Status For the visit The number of the target star, time The amount of control Contains the target star number for the next visit And the number of phase adjustment cycles , expressed as .

9. The method according to claim 7, characterized in that: The speed increment and transfer time of the current round of maneuvers are obtained according to the phase adjustment dynamic calculation method. The cost of this level is represented as a dynamic function in combination with the weighted coefficient input by the program, including: According to the phase adjustment dynamic calculation method, the speed increment and transfer time of the current round of maneuvers are obtained. Combined with the weighted coefficient input by the program, the cost of this level is represented as a dynamic function: in, Indicates the target star number where the current service star is located, determined according to the state quantity. Indicates the number of the next target star to be visited by the service star determined by the control amount. Indicates the number of phase adjustment cycles.

Images